Hans Noel tries to answer the question:
Members of Congress are not polarized because voters are now better sorted. And voters are not polarized simply because legislators now are. The missing piece is ideological activists, who now dominate the political parties. In short, policy demanders. These politically engaged activists are the base that legislators are increasingly playing to, because they are the ones who provide campaign resources and who threaten primary challenges. Their polarization also filters to voters, through elected officials but also through the media and informal networks. (And ultimately, these activists themselves may be polarized because elite political thinkers are polarized, but you don’t have to buy that story to believe that activists are important.) Of course, studying legislative and mass polarization is very important, but its far from the center of the story.
Seth Masket adds that almost no one “gets into politics with the goal of driving the parties further apart.” Instead, he argues, individuals “get involved in politics usually because they want the government to do something different from what it’s currently doing”:
Activists have become better at this over time. They’re increasingly organizing over a broader range of issues and they’ve become adept at getting political parties to adopt their stances, making it even harder for politicians to resist them. This isn’t necessarily a bad thing. Indeed, this is how governing ideas are generated and translated into law. But it’s important to remember that the parties aren’t far apart because people hate each other; they’re far apart because people want the government to do things. This is why exhortations for common ground tend to fall on deaf ears. People favor compromise in principle, except on the one thing that drove them into politics in the first place.
Julia Azari partially blames growing polarization on growing distrust of government:
[P]artisanship and declining trust in government have become mutually reinforcing. In my research, I find that mistrust of governing institutions (I focus on the presidency, although I think we can all agree that Congress has not been immune to this) emerged around the same time that the parties began to sort ideologically in response to the collapse of the New Deal coalition and the rise of cultural issues on the agenda. These began – in the late 1960s – as distinct phenomena. But as time went on, they became intertwined. A general lack of reverence and respect for the office of the presidency – not without good reason after Watergate and Vietnam – have merged with party polarization to create an environment in which presidents tend to be divisive, rather than uniting figures. They also tend, as I argue in the book, to rely more on language that appeals to their supporters and their campaign promises, which does little to alleviate the problem. In turn, these developments shape the incentives of individual members of Congress, who have increasingly little reason to collaborate across party lines.